Diagnosing a Weekday vs Weekend Gap in Ad Creative Submission Rates

An advertising platform's self-serve ads-management tool has a creative submission funnel: an advertiser starts building an ad creative and finishes by submitting it. The team has found that the creative submission rate has been consistently higher on weekdays than on weekends for a long time. Analyze the possible causes of the gap and its business impact, then answer a follow-up about what stratifying by advertiser type reveals. ### Clarifying Questions - How is the submission rate defined: submissions divided by creatives started, by sessions in the tool, or by active advertisers? Is a submission credited to the day its creative was started or to the day it was submitted? - Which time zone decides whether an event falls on a weekday or a weekend: one reporting time zone such as UTC, or the advertiser's local time? - How large is the gap, over how many weeks has it been stable, and does it appear in every market? - Which advertiser types use the tool (self-serve small businesses, agencies, large managed accounts), and can creatives also be submitted outside the tool, for example through an API? ### Part 1 — Diagnose the gap Explain how you would find the cause of the weekday–weekend difference, starting with whether the difference is real. ```hint Check the ruler first Ask whether a Saturday covers the same 24 hours for an advertiser in Los Angeles and one in Tokyo, and how a creative drafted on Saturday and submitted on Monday is counted under the metric's definition. ``` ```hint Who shows up on weekends Consider whether the advertisers building creatives on a Saturday are the same kind of advertisers building them on a Tuesday. ``` #### What This Part Should Cover - Validation of the metric definition, logging completeness and the time zone used for day boundaries - Testable hypotheses about advertiser activity, site performance, advertiser size, market, campaign type, and new versus existing advertisers - The funnel step where the weekend drop happens and the segments that drive it ### Part 2 — Lost submissions or postponed ones? Determine whether the lower weekend rate is a real loss of submissions or whether advertisers simply postpone submitting until the working week, and describe the business impact in each case. ```hint Follow the draft Ask what eventually happens to a creative that was started on a Saturday but not submitted that day. ``` ```hint A delay has a price Even a postponed submission costs something. Think about what happens to a campaign while its creative waits. ``` #### What This Part Should Cover - A creative-level or advertiser-level cohort measurement that separates abandonment from deferral - Impact sizing for both cases: submissions, delayed or lost ad spend, and the advertisers affected - Whether the gap deserves product investment, and in which segment ### Part 3 — The gap disappears within advertiser types Follow-up: after you stratify by advertiser type, the weekday and weekend submission rates are about the same within every type. How do you explain the overall gap, and how should it be reported? ```hint Weights and rates An overall rate is a weighted average of segment rates. Ask which of the two differs between weekdays and weekends. ``` #### What This Part Should Cover - The mechanism that produces an overall gap with no gap inside any type - Reporting raw and mix-adjusted rates side by side - What the finding changes about where to look for friction and what to fix ### What a Strong Answer Covers - Validation of the metric before any behavioral explanation - Hypotheses each tied to a specific cut or data source rather than a generic checklist - Cohort tracking that separates lower conversion from shifted timing - A decomposition of the gap into a composition effect and a within-segment rate effect - Impact stated in submissions and ad spend for the affected segments, ending in a recommendation ### Follow-up Questions - Weekend sessions show more upload errors. How would you show that the errors cause abandonment rather than merely occurring alongside it? - If a real within-type gap exists only for new small advertisers, what product change would you test, and how would you design the experiment? - How would your approach change if the gap had appeared suddenly a month ago instead of being long-standing? - Stratifying by advertiser type removes the gap, but stratifying by type and market together shows a weekend gap in a few markets. How do you decide whether that gap is real?